Image Denoising Based On Multi-Scale Analysis

With the development of technology,image has become an important source for everyone to get information for people in modern information age.But in reality,the image will be polluted by the noise inevitably.So how to denoising the image effectively has always been a hot spot in the field of image process.On the basis of introducing the usual image denoising methods, some new denoising algorithm is proposed in this paper,the main innovations are as follows:(1) According to the two important factors of affecting the wavelet threshold denoising effects,the threshold functions and threshold is studied.A new thresholding function that the parameters can be adjusted is proposed in this paper. And then propose a calculation formula which can change by the wavelet decomposition scale and automatic adjustment, to implement process with different threshold size in diverse decomposition.(2) An improved anisotropic diffusion model is proposed to overcome the shortcoming of traditional P-M anisotropic diffusion model, the improved diffusion model add a good edge threshod that can better detect the edge detail of the image.That can better distinguish the edge of image and noise.(3) Based on introducing the basic principle and advantages of the Nonsubsampled contourlet transform(NSCT) and Total Variation(TV) model.The two different types of image processing methods,namely NSCT and partial differential equation,which are combined reasonably. That is to say,the low-frequency component we get after NSCT decomposition could use the TV model to deal with;and for the high-frequency component,we adapt the improved anisotropic diffusion model which metions in the fourth chapter to process,then carrying out NSCT inverse transform for the processed high-frequency components and the low-frequency components.Finally,we obtained an image after denoising.In this paper,a large number of experimental simulation show that using the algorithm of denoising image we proposed has a better visual effect and more details. And analyzing different denoising algorithm under the objective evaluation index,which show that this algorithm has better ability to retain the image edge and texture information. Last but not the least,the important conclusion after analyzing has an important practical guiding significance for the subsequent image processing work.